Queen and pawn endgame

[Maxwell86]

No trolling from my part anyway...

I do believe that there is a big difference between 35....Qxf3 and 35....Qg2+. The former leads to a pawn ending that is a relatively easy win for white. While for latter, black has very good chances for a draw.

After 35....Qg2+, the critical line seems to be 36.Ke3 Qg1+ 37.Ke4 Qe1+ 38.Qe3 Qb4+ 39.Kd3 Qb5+ 40.Kd2 Qxb2 41.Qe8+ Kh7 42.Qxf7 Qd4+ 43.Ke2 Qe5+ 44.Kd3 Qd6+

[d]8/1p3Qpk/p2q3p/7P/6P1/3K1P2/P1P5/8 w - - 5 45

Sure, black is a pawn down, but he can keep harassing the white king. I would guess it's a draw.

[Lyudmil Tsvetkov]

No trolling from my part anyway...

I do believe that there is a big difference between 35....Qxf3 and 35....Qg2+. The former leads to a pawn ending that is a relatively easy win for white. While for latter, black has very good chances for a draw.

After 35....Qg2+, the critical line seems to be 36.Ke3 Qg1+ 37.Ke4 Qe1+ 38.Qe3 Qb4+ 39.Kd3 Qb5+ 40.Kd2 Qxb2 41.Qe8+ Kh7 42.Qxf7 Qd4+ 43.Ke2 Qe5+ 44.Kd3 Qd6+

[d]8/1p3Qpk/p2q3p/7P/6P1/3K1P2/P1P5/8 w - - 5 45

Sure, black is a pawn down, but he can keep harassing the white king. I would guess it's a draw.

Hi Bart.

nice to meet you.

I am certain you have analysed your game thoroughly.

right, queen ending gives more chances to practically try to draw.

I suppose you also picked the right position to analyse(we are all using engines to assist us, are not we?), however, my short attempts with SF and Komodo consistently lead to white scores only increasing.

I reached +150cps from both engines and stopped there.

usually, consistently increasing scores would denote a side's win, but you never know what tbs oddity might pop up from here and there.

in order to spark a discussion(what is this forum otherwise for?), my assessment would be: white wins.

my eval assessment also tells me so.

maybe our respected huge-machine contributors from all over the world can clear things up a bit.

PS. oh, no, Louis, not again 0.0, tbs draw.

[zullil]

No trolling from my part anyway...

I do believe that there is a big difference between 35....Qxf3 and 35....Qg2+. The former leads to a pawn ending that is a relatively easy win for white. While for latter, black has very good chances for a draw.

After 35....Qg2+, the critical line seems to be 36.Ke3 Qg1+ 37.Ke4 Qe1+ 38.Qe3 Qb4+ 39.Kd3 Qb5+ 40.Kd2 Qxb2 41.Qe8+ Kh7 42.Qxf7 Qd4+ 43.Ke2 Qe5+ 44.Kd3 Qd6+

[d]8/1p3Qpk/p2q3p/7P/6P1/3K1P2/P1P5/8 w - - 5 45

Sure, black is a pawn down, but he can keep harassing the white king. I would guess it's a draw.

Hi Bart.

nice to meet you.

I am certain you have analysed your game thoroughly.

right, queen ending gives more chances to practically try to draw.

I suppose you also picked the right position to analyse(we are all using engines to assist us, are not we?), however, my short attempts with SF and Komodo consistently lead to white scores only increasing.

I reached +150cps from both engines and stopped there.

usually, consistently increasing scores would denote a side's win, but you never know what tbs oddity might pop up from here and there.

in order to spark a discussion(what is this forum otherwise for?), my assessment would be: white wins.

my eval assessment also tells me so.

maybe our respected huge-machine contributors from all over the world can clear things up a bit.

PS. oh, no, Louis, not again 0.0, tbs draw.

Here is what asmFIsh with 6-man tables gives after a long search on the initial position.
[d] 6k1/1p3pp1/p6p/7P/3Q2P1/5P2/PPPK4/5q2 b - - 2 35

Not 0.00, but no progress for White at all. The evaluation has remained constant for many iterations. So I conclude that Bart is correct and Lyudmil is not.

Code: Select all

+0.96  35... Qg2+ 36. Ke3 Qg1+ 37. Ke4 Qe1+ 38. Qe3 Qb4+ 39. Kd3 Qb5+ 40. Kd2 Qxb2 41. Qe8+ Kh7 42. Qxf7 Qd4+ 43. Ke2 Qe5+ 44. Kd3 Qd6+ 45. Kc3 Qe5+ 46. Kc4 b5+ 47. Kb4 Qe1+ 48. c3 Qd2 49. Ka5 Qxc3+ 50. Kxa6 Qc4 51. Qg6+ Kh8 52. Qe8+ Kh7 53. Qe4+ Kg8 54. Qe3 Qa4+ 55. Kb6 Qxa2 56. Kxb5 Kh8 57. Qe4 Qb3+ 58. Kc5 Qc3+ 59. Kd5 Qd2+ 60. Kc4 Qc1+ 61. Kd3 Qd1+ 62. Ke3 Qe1+ 63. Kf4 Qc1+ 64. Qe3 Qc7+ 65. Kf5 Qc2+ 66. Ke5 Qb2+ 67. Ke4 Qc2+ 68. Kd5 Qa2+ 69. Kd4 Qa4+ 70. Kd3 Qd1+ 71. Kc4 Qa4+ 72. Kc3 Qd1 73. Qf4 Qe1+ 74. Qd2 Qa1+ 75. Kd3 Qb1+ 76. Ke2 Qb5+ 77. Kf2 (depth 65, 3:05:14)

[MikeGL]

I wonder, is this common for pawn endings?
Some pawn races can be quite deep, which might explain why an engine needs some time to calculate them. But at the same time, these lines are quite forced, which would limit the branches of the search.

If you want to see a really complicated pawn ending for engines, try opening my thread "An easy position, for humans". Then, perform a search on that position....

I noticed there's extra pawn in there too, but only ends up in draw because of both Q still on the board.

Maybe this is a powerful engine eval idea which is being opened up by GM Michiels.
For engines to choose a pawn down endgame with both Q on the board, versus a purely pawn endgame (all pawns equal) with opposite king activated/centralized.
Ok I agree, very deep search compensates and shows the exact route (Qg2+!) for the draw, but on blitz or lightning, engines would miss this idea of Qg2+! which saves, against the weak Qxf3?! pawn capture.
I remember the Bednorz setting of TheKing (also called ChessMaster) where it sets all of its own pieces with +0.3 value which encourages complications and not to trade pieces unnecesarily. Bednorz setting was stronger than the default ChessMaster setting. That Bednorz setting of TheKing would surely choose Qg2+ on this position right away compared to Qxf3.

regards

[Lyudmil Tsvetkov]

I wonder, is this common for pawn endings?
Some pawn races can be quite deep, which might explain why an engine needs some time to calculate them. But at the same time, these lines are quite forced, which would limit the branches of the search.

If you want to see a really complicated pawn ending for engines, try opening my thread "An easy position, for humans". Then, perform a search on that position....

I noticed there's extra pawn in there too, but only ends up in draw because of both Q still on the board.

Maybe this is a powerful engine eval idea which is being opened up by GM Michiels.
For engines to choose a pawn down endgame with both Q on the board, versus a purely pawn endgame (all pawns equal) with opposite king activated/centralized.
Ok I agree, very deep search compensates and shows the exact route (Qg2+!) for the draw, but on blitz or lightning, engines would miss this idea of Qg2+! which saves, against the weak Qxf3?! pawn capture.
I remember the Bednorz setting of TheKing (also called ChessMaster) where it sets all of its own pieces with +0.3 value which encourages complications and not to trade pieces unnecesarily. Bednorz setting was stronger than the default ChessMaster setting. That Bednorz setting of TheKing would surely choose Qg2+ on this position right away compared to Qxf3.

regards

no clear draw here, as well.

rather, I would say, white wins the queen ending too.

but choosing it might still be the wiser way to go.

[Lyudmil Tsvetkov]

No trolling from my part anyway...

I do believe that there is a big difference between 35....Qxf3 and 35....Qg2+. The former leads to a pawn ending that is a relatively easy win for white. While for latter, black has very good chances for a draw.

After 35....Qg2+, the critical line seems to be 36.Ke3 Qg1+ 37.Ke4 Qe1+ 38.Qe3 Qb4+ 39.Kd3 Qb5+ 40.Kd2 Qxb2 41.Qe8+ Kh7 42.Qxf7 Qd4+ 43.Ke2 Qe5+ 44.Kd3 Qd6+

[d]8/1p3Qpk/p2q3p/7P/6P1/3K1P2/P1P5/8 w - - 5 45

Sure, black is a pawn down, but he can keep harassing the white king. I would guess it's a draw.

Hi Bart.

nice to meet you.

I am certain you have analysed your game thoroughly.

right, queen ending gives more chances to practically try to draw.

I suppose you also picked the right position to analyse(we are all using engines to assist us, are not we?), however, my short attempts with SF and Komodo consistently lead to white scores only increasing.

I reached +150cps from both engines and stopped there.

usually, consistently increasing scores would denote a side's win, but you never know what tbs oddity might pop up from here and there.

in order to spark a discussion(what is this forum otherwise for?), my assessment would be: white wins.

my eval assessment also tells me so.

maybe our respected huge-machine contributors from all over the world can clear things up a bit.

PS. oh, no, Louis, not again 0.0, tbs draw.

Here is what asmFIsh with 6-man tables gives after a long search on the initial position.
[d] 6k1/1p3pp1/p6p/7P/3Q2P1/5P2/PPPK4/5q2 b - - 2 35

Not 0.00, but no progress for White at all. The evaluation has remained constant for many iterations. So I conclude that Bart is correct and Lyudmil is not.

Code: Select all

+0.96  35... Qg2+ 36. Ke3 Qg1+ 37. Ke4 Qe1+ 38. Qe3 Qb4+ 39. Kd3 Qb5+ 40. Kd2 Qxb2 41. Qe8+ Kh7 42. Qxf7 Qd4+ 43. Ke2 Qe5+ 44. Kd3 Qd6+ 45. Kc3 Qe5+ 46. Kc4 b5+ 47. Kb4 Qe1+ 48. c3 Qd2 49. Ka5 Qxc3+ 50. Kxa6 Qc4 51. Qg6+ Kh8 52. Qe8+ Kh7 53. Qe4+ Kg8 54. Qe3 Qa4+ 55. Kb6 Qxa2 56. Kxb5 Kh8 57. Qe4 Qb3+ 58. Kc5 Qc3+ 59. Kd5 Qd2+ 60. Kc4 Qc1+ 61. Kd3 Qd1+ 62. Ke3 Qe1+ 63. Kf4 Qc1+ 64. Qe3 Qc7+ 65. Kf5 Qc2+ 66. Ke5 Qb2+ 67. Ke4 Qc2+ 68. Kd5 Qa2+ 69. Kd4 Qa4+ 70. Kd3 Qd1+ 71. Kc4 Qa4+ 72. Kc3 Qd1 73. Qf4 Qe1+ 74. Qd2 Qa1+ 75. Kd3 Qb1+ 76. Ke2 Qb5+ 77. Kf2 (depth 65, 3:05:14)

and if you repeat that one from the position Bart posted, to help your search horizon?

SF should not be changing all pawns on the queen side here, rather go for an f3-f4-f5 storm, something like that.

my 3 minute-long quick experiments with Komodo and SF lead to increasing score with both engines in most of the lines.

but again, my engines are, of course, buggy.

[MikeGL]

I wonder, is this common for pawn endings?
Some pawn races can be quite deep, which might explain why an engine needs some time to calculate them. But at the same time, these lines are quite forced, which would limit the branches of the search.

If you want to see a really complicated pawn ending for engines, try opening my thread "An easy position, for humans". Then, perform a search on that position....

I noticed there's extra pawn in there too, but only ends up in draw because of both Q still on the board.

Maybe this is a powerful engine eval idea which is being opened up by GM Michiels.
For engines to choose a pawn down endgame with both Q on the board, versus a purely pawn endgame (all pawns equal) with opposite king activated/centralized.
Ok I agree, very deep search compensates and shows the exact route (Qg2+!) for the draw, but on blitz or lightning, engines would miss this idea of Qg2+! which saves, against the weak Qxf3?! pawn capture.
I remember the Bednorz setting of TheKing (also called ChessMaster) where it sets all of its own pieces with +0.3 value which encourages complications and not to trade pieces unnecesarily. Bednorz setting was stronger than the default ChessMaster setting. That Bednorz setting of TheKing would surely choose Qg2+ on this position right away compared to Qxf3.

regards

no clear draw here, as well.

rather, I would say, white wins the queen ending too.

but choosing it might still be the wiser way to go.

Sorry, ; ) but with engine analysis posted above plus the conclusions of GM Gurevich and the original poster GM Michiels, I think I would also conclude that Qxf3 is a +- and the move Qg2+ only =

Below is a line of L.Zulli inserted with a PGN header
[pgn]
[Event "Chess Olympiad"]
[Site "Khanty-Mansiysk RUS"]
[Date "2010.09.24"]
[Round "4"]
[White "Leinier Dominguez Perez"]
[Black "Bart Michiels"]
[Result "1/2-1/2"]
[BlackElo "2459"]
[WhiteElo "2716"]
[TimeControl "60"]
[SetUp "1"]
[FEN "6k1/1p3pp1/p6p/7P/3Q2P1/5P2/PPPK4/5q2 b - - 0 35"]
[Termination "normal"]
[PlyCount "84"]
[WhiteType "program"]
[BlackType "human"]

35. ... Qg2+ 36. Ke3 Qg1+ 37. Ke4 Qe1+ 38. Qe3 Qb4+ 39. Kd3 Qb5+ 40. Kd2
Qxb2 41. Qe8+ Kh7 42. Qxf7 Qd4+ 43. Ke2 Qe5+ 44. Kd3 Qd6+ 45. Kc3 Qe5+ 46.
Kc4 b5+ 47. Kb4 Qe1+ 48. c3 Qd2 49. Ka5 Qxc3+ 50. Kxa6 Qc4 51. Qg6+ Kh8 52.
Qe8+ Kh7 53. Qe4+ Kg8 54. Qe3 Qa4+ 55. Kb6 Qxa2 56. Kxb5 Kh8 57. Qe4 Qb3+
58. Kc5 Qc3+ 59. Kd5 Qd2+ 60. Kc4 Qc1+ 61. Kd3 Qd1+ 62. Ke3 Qe1+ 63. Kf4
Qc1+ 64. Qe3 Qc7+ 65. Kf5 Qc2+ 66. Ke5 Qb2+ 67. Ke4 Qc2+ 68. Kd5 Qa2+ 69.
Kd4 Qa4+ 70. Kd3 Qd1+ 71. Kc4 Qa4+ 72. Kc3 Qd1 73. Qf4 Qe1+ 74. Qd2 Qa1+
75. Kd3 Qb1+ 76. Ke2 Qb5+ 77. Kf2 1/2-1/2
[/pgn]

Even endgame tablebases would agree if you setup any KQp vs KQ where the extra pawn won't make a win, except on very few special cases.

regards

[Lyudmil Tsvetkov]

quick shootout with SF and Komodo, 10 games each, 1 min. per game, from the position Bart posted:

[d]8/1p3Qpk/p2q3p/7P/6P1/3K1P2/P1P5/8 w - - 0 1

12 white wins, 8 draws

SF wins half, Komodo 7

conclude from that what you want.

2 white wins:

[pgn][Event "Shootout (Stockfish864POPCNT, Blitz 1m)"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "New game"]
[Black "?"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/1p3Qpk/p2q3p/7P/6P1/3K1P2/P1P5/8 w - - 0 1"]
[PlyCount "111"]

1. Kc3 {0.53/18 1} Qe5+ {0.81/20 2} 2. Kc4 {1.14/21 2} Qb5+ {1.07/22 3} 3. Kd4
{1.10/21 1} Qb6+ {1.10/23 2} 4. Ke5 {1.14/23 8 (Kc3)} Qb2+ {1.06/22 3 (Qe3+)}
5. Kd6 {1.20/22 1} Qb6+ {1.26/22 3} 6. Kd7 {1.28/23 1} Qc6+ {1.23/22 2} 7. Kd8
{1.21/23 1} b5 {1.31/23 2} 8. Qf5+ {1.33/22 1} Kh8 {1.28/23 2} 9. Qe4 {1.27/21
0} Qf6+ {1.10/22 3} 10. Kc7 {1.08/23 0} Qf7+ {1.15/24 2} 11. Kb6 {1.15/24 1}
Qxa2 {1.17/24 2} 12. f4 {1.19/24 1} Qg8 {1.19/25 2} 13. f5 {1.19/25 1} b4 {1.
04/24 3} 14. Qe6 {1.12/21 1} Qa8 {1.25/23 3} 15. f6 {1.26/22 0} Qd8+ {1.25/22 1
} 16. Kxa6 {1.24/24 1} gxf6 {1.25/24 1} 17. Qf5 {1.25/24 0} Qd4 {1.26/23 1} 18.
Kb5 {1.26/25 2} Qd6 {1.25/23 1 (Qb2)} 19. Qe4 {1.26/21 1 (Ka4)} Qb8+ {1.26/22 1
} 20. Kc4 {1.24/22 1} b3 {1.17/22 1} 21. cxb3 {1.22/23 1} Qc7+ {1.16/23 1} 22.
Kd3 {1.57/24 7} Qg3+ {1.59/20 1} 23. Qe3 {1.71/20 1} Qd6+ {1.77/21 2} 24. Qd4 {
1.67/22 1} Qg3+ {1.82/19 1} 25. Kc4 {1.82/21 1} Qc7+ {1.86/19 1} 26. Qc5 {1.85/
22 1} Qf4+ {1.86/21 1} 27. Kb5 {1.93/23 1} Qxg4 {2.01/21 2} 28. Qf8+ {2.16/21 0
} Kh7 {2.16/22 1} 29. Qf7+ {2.47/21 1} Kh8 {2.47/22 0} 30. Qxf6+ {2.45/23 1}
Kh7 {2.54/22 0} 31. Qe7+ {2.58/23 1} Kh8 {2.61/19 0 (Kg8)} 32. Qe5+ {2.41/20 1}
Kg8 {2.44/18 0} 33. Qd5+ {2.61/19 1} Kh8 {2.69/21 1} 34. b4 {2.69/20 0} Qg7 {
2.68/20 1} 35. Qc6 {2.73/17 0} Qg5+ {2.67/19 1 (Qe5+)} 36. Ka6 {2.78/19 0} Qxh5
{2.72/20 0} 37. b5 {2.83/19 0} Qe2 {2.83/19 0} 38. Qxh6+ {3.46/19 2} Kg8 {3.46/
1 0} 39. Qd6 {3.52/22 0} Kf7 {3.52/19 0} 40. Kb6 {3.59/23 0} Ke8 {3.66/23 1}
41. Qd5 {3.66/25 0} Ke7 {3.66/23 0 (Qb2)} 42. Kc6 {3.66/22 0} Qg4 {3.40/19 0}
43. Qd6+ {3.29/20 1} Ke8 {3.41/21 0} 44. b6 {3.58/21 0} Qc4+ {3.66/19 0 (Qf3+)}
45. Qc5 {3.41/21 0} Qe4+ {3.58/21 0 (Qe6+)} 46. Kc7 {3.66/23 0 (Qd5)} Qf4+ {3.
35/21 0 (Qh7+)} 47. Kc8 {2.72/21 1 (Qd6)} Qg4+ {3.35/21 0} 48. Kb7 {3.66/21 0}
Qh4 {3.74/20 0} 49. Qd6 {3.61/23 0} Qf2 {3.58/21 0} 50. Kb8 {3.74/21 0} Qb2 {
3.74/22 0 (Qa2)} 51. b7 {3.74/25 0} Qa1 {3.91/22 0 (Qa2)} 52. Qe6+ {7.25/21 1
(Qc5)} Kd8 {3.27/17 0} 53. Qd5+ {2.80/19 0} Ke7 {2.72/13 0} 54. Kc7 {2.50/18 0}
Qc3+ {2.56/12 0 (Qc1+)} 55. Kb6 {7.26/19 1} Qb2+ {7.72/13 0 (Qb4+)} 56. Kc6 {
10.75/15 0} 1-0

[Event "Shootout (Komodo1164-bit, Blitz 1m) "]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "New game"]
[Black "?"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/1p3Qpk/p2q3p/7P/6P1/3K1P2/P1P5/8 w - - 0 1"]
[PlyCount "117"]

1. Kc3 {0.45/18 1} Qe5+ {0.31/21 2} 2. Kd3 {0.43/22 3 (Kb3)} Qd6+ {0.52/20 1}
3. Kc4 {0.54/21 3} Qc6+ {0.63/20 1} 4. Kb3 {0.56/22 1} Qb5+ {0.68/22 4} 5. Ka3
{0.56/19 0} Qc5+ {0.70/20 1} 6. Kb2 {0.71/21 1} Qd4+ {0.72/20 1} 7. Kb3 {0.56/
21 0} Qe3+ {0.57/21 1} 8. Kc4 {0.69/21 1} Qe2+ {0.71/20 1} 9. Kd5 {0.73/20 1}
Qb5+ {0.73/21 2} 10. Kd4 {0.80/21 2} Qb6+ {0.71/21 1} 11. Ke5 {0.77/21 1} Qb2+
{0.74/21 1} 12. Kd6 {0.73/21 1} Qb6+ {0.71/22 1} 13. Kd7 {0.71/22 1} Qc6+ {0.
67/21 1} 14. Kd8 {0.74/21 1} b5 {0.75/21 1} 15. Qf5+ {0.79/22 2} Kh8 {0.80/20 1
} 16. Qc8 {0.92/21 2} Qb6+ {0.92/20 2 (Qd6+)} 17. Kd7+ {1.00/20 1} Kh7 {1.00/
20 0} 18. Qc6 {1.00/19 0} Qa7+ {0.97/21 1} 19. Kd6 {1.00/21 1} Qd4+ {1.00/21 2}
20. Qd5 {1.00/21 1} Qb6+ {1.02/21 1} 21. Kd7 {0.93/21 1} b4 {0.93/21 2} 22. Qe6
{1.02/20 1} Qb7+ {1.02/20 1} 23. Kd6 {1.02/21 1} Qb6+ {1.02/21 1} 24. Kd5 {1.
03/21 1} Qc7 {0.96/21 2 (Qa5+)} 25. Qxa6 {1.68/19 2} Kh8 {1.67/18 1 (Qxc2)} 26.
Qe6 {1.69/21 1} Qa5+ {1.67/22 3 (Qc3)} 27. Kd4 {1.74/23 2} Qa7+ {1.74/23 1} 28.
Kd3 {1.74/24 1} Qa3+ {1.74/23 1 (Qc5)} 29. Qb3 {1.74/19 0} Qa6+ {1.74/21 1} 30.
Qc4 {1.72/22 2} Qd6+ {1.76/23 1} 31. Qd4 {1.76/23 0} Qa6+ {1.76/24 1} 32. Kd2 {
1.77/22 1} Qa5 {1.78/22 1} 33. Kc1 {1.79/22 0 (Kd1)} Qxa2 {1.82/22 0} 34. Qxb4
{1.82/21 0} Qa8 {1.84/21 1 (Qa7)} 35. Qe4 {1.96/21 1} Qa3+ {2.04/21 1} 36. Kd1
{2.16/22 2} Qc3 {2.16/22 1 (Qd6+)} 37. Qd5 {2.18/23 1} Qa1+ {2.18/22 0} 38. Ke2
{2.18/22 0} Qb2 {2.18/23 1 (Qg1)} 39. Qd8+ {2.19/23 1} Kh7 {2.12/5 0} 40. Qd2 {
2.20/21 0 (Qd3+)} Qe5+ {2.22/22 1} 41. Kf1 {2.23/22 0} Qb2 {2.23/20 0} 42. Qd3+
{2.54/22 1} Kh8 {2.55/22 0} 43. Qd8+ {2.55/22 0} Kh7 {2.55/5 0} 44. c4 {2.55/
22 0} Qc1+ {2.55/23 0} 45. Ke2 {2.67/24 2} Qb2+ {2.67/21 0 (Qg1)} 46. Qd2 {2.
67/19 0} Qe5+ {2.68/18 0 (Qb6)} 47. Qe3 {2.68/20 0} Qb2+ {2.68/21 1} 48. Kf1 {
2.71/21 1} Qh2 {2.71/20 0} 49. c5 {2.83/20 1} Qh1+ {2.77/17 0} 50. Ke2 {2.89/
18 0} Qh2+ {2.89/16 0} 51. Kd3 {2.89/18 0} Qb8 {2.95/19 1} 52. c6 {2.96/19 0}
Kh8 {3.08/19 1} 53. Qe6 {3.24/17 0} Qb1+ {3.30/17 0} 54. Kd4 {3.71/16 0} Qb4+ {
3.82/17 1 (Qb8)} 55. Kd5 {4.03/19 0} Qb8 {4.75/17 1} 56. Qd7 {4.90/17 0} Qb3+ {
5.77/18 1} 57. Kc5 {5.92/19 0} Qe3+ {6.73/19 1} 58. Kb5 {7.22/17 0} Qb3+ {7.83/
19 1} 59. Ka6 {7.84/17 0} 1-0

[/pgn]

[Lyudmil Tsvetkov]

I wonder, is this common for pawn endings?
Some pawn races can be quite deep, which might explain why an engine needs some time to calculate them. But at the same time, these lines are quite forced, which would limit the branches of the search.

If you want to see a really complicated pawn ending for engines, try opening my thread "An easy position, for humans". Then, perform a search on that position....

I noticed there's extra pawn in there too, but only ends up in draw because of both Q still on the board.

Maybe this is a powerful engine eval idea which is being opened up by GM Michiels.
For engines to choose a pawn down endgame with both Q on the board, versus a purely pawn endgame (all pawns equal) with opposite king activated/centralized.
Ok I agree, very deep search compensates and shows the exact route (Qg2+!) for the draw, but on blitz or lightning, engines would miss this idea of Qg2+! which saves, against the weak Qxf3?! pawn capture.
I remember the Bednorz setting of TheKing (also called ChessMaster) where it sets all of its own pieces with +0.3 value which encourages complications and not to trade pieces unnecesarily. Bednorz setting was stronger than the default ChessMaster setting. That Bednorz setting of TheKing would surely choose Qg2+ on this position right away compared to Qxf3.

regards

no clear draw here, as well.

rather, I would say, white wins the queen ending too.

but choosing it might still be the wiser way to go.

Sorry, ; ) but with engine analysis posted above plus the conclusions of GM Gurevich and the original poster GM Michiels, I think I would also conclude that Qxf3 is a +- and the move Qg2+ only =

Below is a line of L.Zulli inserted with a PGN header
[pgn]
[Event "Chess Olympiad"]
[Site "Khanty-Mansiysk RUS"]
[Date "2010.09.24"]
[Round "4"]
[White "Leinier Dominguez Perez"]
[Black "Bart Michiels"]
[Result "1/2-1/2"]
[BlackElo "2459"]
[WhiteElo "2716"]
[TimeControl "60"]
[SetUp "1"]
[FEN "6k1/1p3pp1/p6p/7P/3Q2P1/5P2/PPPK4/5q2 b - - 0 35"]
[Termination "normal"]
[PlyCount "84"]
[WhiteType "program"]
[BlackType "human"]

35. ... Qg2+ 36. Ke3 Qg1+ 37. Ke4 Qe1+ 38. Qe3 Qb4+ 39. Kd3 Qb5+ 40. Kd2
Qxb2 41. Qe8+ Kh7 42. Qxf7 Qd4+ 43. Ke2 Qe5+ 44. Kd3 Qd6+ 45. Kc3 Qe5+ 46.
Kc4 b5+ 47. Kb4 Qe1+ 48. c3 Qd2 49. Ka5 Qxc3+ 50. Kxa6 Qc4 51. Qg6+ Kh8 52.
Qe8+ Kh7 53. Qe4+ Kg8 54. Qe3 Qa4+ 55. Kb6 Qxa2 56. Kxb5 Kh8 57. Qe4 Qb3+
58. Kc5 Qc3+ 59. Kd5 Qd2+ 60. Kc4 Qc1+ 61. Kd3 Qd1+ 62. Ke3 Qe1+ 63. Kf4
Qc1+ 64. Qe3 Qc7+ 65. Kf5 Qc2+ 66. Ke5 Qb2+ 67. Ke4 Qc2+ 68. Kd5 Qa2+ 69.
Kd4 Qa4+ 70. Kd3 Qd1+ 71. Kc4 Qa4+ 72. Kc3 Qd1 73. Qf4 Qe1+ 74. Qd2 Qa1+
75. Kd3 Qb1+ 76. Ke2 Qb5+ 77. Kf2 1/2-1/2
[/pgn]

Even endgame tablebases would agree if you setup any KQp vs KQ where the extra pawn won't make a win, except on very few special cases.

regards

man, don't you have some shopping to do for someone else?

ok, too many GMs here, I feel a bit uncomfortable.

Louis Zulli line is second or 3rd-rated one, I guess his SF will find better with shorter game horizon.

on the contrary, most QP vs Q endgames(possibly couple more pawns on the other flank) are wins for the stronger side; that would be true, if the stronger side has any pawn apart from edge a or h ones, so b,c,d ones would do just fine.

see also the games I posted below: 12 white wins from 20 games with SF and Komodo.

[Maxwell86]

Sorry, ; ) but with engine analysis posted above plus the conclusions of GM Gurevich and the original poster GM Michiels, I think I would also conclude that Qxf3 is a +- and the move Qg2+ only =

Well, Gurevich and I didn't go very deeply to find out whether the queen ending is a draw or not. It was more about the mistake in my annotations, that 35....Qxf3 loses immediately and that the move I played was the best chance after all.